cell count

cell count  

  By: tanbobo on Nov. 21, 2021, 8:01 a.m.

As the example image shown in the Data Format, the number of cell described in the counts.csv is not the same with the mask image. For instance, from the counts.csv, the count of the first cell type called neutrophil in image[3333] is 2, but, after visualization of this image, for neutrophil, the mask number is 3.

Re: cell count  

  By: simongraham73 on Nov. 21, 2021, 8:46 a.m.

Hi!

If you look just below the image examples, you will see the following:

"A given nucleus is considered present in the image if any part of it is within the central 224x224 region within the patch. This ensures that a nucleus is only considered for counting if it lies completely within the original 256x256 image."

Below we show the neutrphils in red and all other nuclei in blue. We also show the central 224x224 region in green. You can see that just two neutrophils lie within the central region.

I hope this clarifies it. We will try and make this information clearer on the website.

Good luck with your experiments.

Simon

Re: cell count  

  By: tanbobo on Nov. 21, 2021, 9:28 a.m.

Oh, yeah, I notice the key sentence upon the Data Download, Thank you for your time.

Re: cell count  

  By: tanbobo on Nov. 21, 2021, 9:43 a.m.

This is a challenging segmentation competition containing instance seg, semantic seg, small objects processing, class imbalance, color imbalance and there are two months to go before making submission. Amazing....